nllb-indo-en

This model is a fine-tuned version of facebook/nllb-200-distilled-600M on Fleurs Dataset without duplication of ids. It achieves the following results on the evaluation set:

• Loss: 0.3048

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	1.0	96	4.3318
53.1274	2.0	192	2.0475
25.7634	3.0	288	0.4936
8.4388	4.0	384	0.2444
1.7896	5.0	480	0.2407
0.8853	6.0	576	0.2626
0.5583	7.0	672	0.2793
0.4353	8.0	768	0.2936
0.3497	9.0	864	0.2992
0.2969	10.0	960	0.3038
0.2713	10.4199	1000	0.3048

Model Evaluation

The performance of this model was evaluated using BLEU and CHRF metrics on validation dataset.

BLEU	CHRF
40.94	66.46

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0